An early study of bacterial vaginosis (BV) involved comparisons of the pH of vaginal fluids of women known to be suffering from BV with those known to be free of the disease. Gardner, H. L., et al., Am. J. Obstet. Gynecol. 69: 962 (1955). All of the BV positive women in the study were determined to have a vaginal fluid pH greater than 4.5, and 91% of these women had a vaginal fluid pH greater than 5.0. Of the normal (disease-free) women in the test, 92% were found to have vaginal pH between 4.0 and 4.7. The conclusion drawn from the study was that a vaginal pH equal to or greater than 5.0 in conjunction with other clinical criteria was indicative of the presence of BV. Subsequent studies culminating in a report by Amsel, R., et al., Am. J. Med. 74: 14-22 (1983), resulted in a reduction of the pH threshold for BV to 4.5, and established the remaining criteria as vaginal fluid homogeneity, the whiff test (treatment with alkali followed by an olfactory test to detect for an amine odor), and the presence of clue cells. These are commonly referred to as the Amsel clinical criteria for BV. The conclusion was based on a study group of 397 women in which 81% of BV positive women were found to have a pH greater than 4.5 while only 23% of the normal women were found to have a vaginal fluid pH greater than 4.5.
Studies subsequent to the report by Amsel et al. have now adjusted the pH threshold to 4.7. One of these is the study of Holst, E., J. Clin. Microbiol. 28: 2035-2039 (1990), in which 100% of the women diagnosed as BV positive by the Amsel criteria were reported to have vaginal fluid pH greater than 4.7. Another is the study by Eschenbach, D. A., Am. J. Obstet. Gynecol. 158(4): 819-828 (1988), in which all 257 women in the study group who had at least 20% clue cells were shown to have a vaginal fluid pH greater than or equal to 4.7, leading to the conclusion that a threshold value of 4.7 correlated best with the other clinical evidence of BV. Krohn, M. A., et al., J. Clin. Microbiol. 27(6): 1266-1271 (1989), also verified the correlation between the vaginal fluid pH threshold of 4.7 and the presence of clue cells, and Holmes, K. K., and coworkers further confirmed the pH 4.7 threshold as an indicator of BV--Holmes, K. K., et al., eds., Sexually Transmitted Diseases, McGraw-Hill, New York (1990), Chapter 46: 527-545 (Holmes, K. K., et al.), and Chapter 47: 547-559 (Hillier, S. L., et al.).
Colorimetric tests for elevated vaginal fluid pH have used nitrazine yellow as an indicator. Nitrazine yellow is a monoazo dye that is bright yellow at pH 6.0 and bright blue at pH 7.2, and has a grey-green midpoint at pH 6.6. In the range of interest for testing vaginal fluid, however, which is approximately 3.5 to 6.0, the change occurs in subtle progressions of grey-green that are difficult to interpret.
pH thresholds are useful indicators in diagnosing a variety of other biological conditions, in both humans and animals, and a large number of colorimetric indicators are known and commercially available. Selection of the appropriate indicator is not always a simple task, however, and the choices are often limited, particularly when a specific color change is desired and when stability of the indicator is a consideration.
Normal bovine milk, for example, has a pH of 6.5 to 6.8, and it has been reported (N. Z. J. Sci. Technol. 27: 258 (1945)) that bovine milk with a pH greater than 6.8 may indicate the presence of bovine mastitis. The difficulties of detecting a pH deviation of such a small magnitude using conventional paper indicator strips are discussed above. A study specifically directed to bovine mastitis was reported in J. Dairy Sci. 68: 1263-1269 (1985). The purpose of the study was to determine the suitability of using absorbent blotting paper impregnated with the pH indicator bromthymol blue to test the pH of bovine milk as a method of detecting subclinical bovine mastitis. Milk was added to the indicator-treated paper, and the color of the pH indicator spot was scored on a scale of 1 to 4, where 1 (pale green) was assessed as normal (negative), and 2, 3 and 4 (increasing from moderate green to dark blue-green) were considered abnormal (positive). The pH of the milk was also determined electronically with a carefully calibrated pH meter. The test results illustrated the difficulty in defining accurately the color of the test area: the predictive value of a positive colorimetric test ranged from 49% to 52% (i.e., 51% to 48% of the results were false positives).
As expected, an increase in the test score was accompanied by an increase in the severity of mastitis as defined by other diagnostic measures. However, in milk from animals with less severe mastitis, the considerable overlapping of results "highlighted the possible error in interpretation of indicator scores. The wide variation of milk pH as determined electronically! within each BTB Brom Thymol Blue color! score showed that the indicator results were not closely related to pH." The investigators stressed the importance of using color comparators which resemble as closely as possible the actual pH test method being utilized. If the colorimetric milk pH test results were to be interpreted immediately, it was important to use comparators that were also wetted with milk. If the colorimetric pH test results were to be determined after the milk spots had dried, it was advantageous to use dry comparators.
Returning to bacterial vaginosis, the whiff test, which is one of the Amsel criteria, originated in a study by Pheifer, et al., N. Engl. J. Med. 298: 1429-1434 (1978), that reported the presence of a characteristic fishy amine odor upon the addition of 10% KOH to a vaginal fluid specimen from a woman with BV. The odor is caused by the alkaline volatilization of amine salts found in the vaginal fluid of women with BV. Unfortunately, the test is highly subjective, it exposes the health care worker to potential biological hazards, and it is disagreeable and vulnerable to error, since it is performed on a microscope slide which, due to the transient nature of the amine odor, must be placed directly under the nose and sniffed immediately after the addition of the KOH.
Alternatives to the whiff test are analytical procedures such as high voltage electrophoresis (Chen, K. C. S., et al., J. Clin. Invest. 63: 828-835 (1979)), thin-layer chromatography (Chen, K. C. S., et al., J. Infect. Dis. 145: 337-345 (1982), and Sanderson, B. E., et al., Br. J. Vener. Dis. 59: 302-305 (1983)), gas chromatography (Gravett, M. G., et al. Obstet. Gynecol. 67: 229-237 (1986), and Dravenieks, A., et al., J. Pharma. Sc. 59: 495-501 (1970)), and high-performance liquid chromatography (Cook, R. L., et al., J. Clin. Microbiol. 30: 870-877 (1992)). These procedures, although more accurate and reliable than the whiff test, are expensive, time-consuming, and not suitable for on-site testing in a physician's office or clinic.
Clue cells, which constitute a further Amsel criterion, are independently correlated with BV, and in the hands of a skilled microscopist are a very sensitive and specific indication of this infection. Clue cells are squamous vaginal epithelial cells found in vaginal fluid when BV is present. The cells are covered with numerous bacteria, giving them a stippled or granular appearance, and their borders are obscured or fuzzy because of the adherence of numerous rods or cocci. According to standard clinical practice, a diagnosis of BV is established when at least 20% of the detectable epithelial cells are clue cells. Holmes, et al., Sexually Transmitted Diseases, 2d ed., McGraw-Hill, Inc., New York, 1990.
Distinguishing between true clue cells in which the adherent bacteria completely obscure the edges of the cells and cells with simply a few adherent bacteria requires training and experience. One source of error is similarity in appearance between clue cells and trichomonads, white blood cells and other vaginal fluid components, frequently resulting in an incorrect identification of these cells as clue cells, and therefore false positive test results. Another is that clue cells when present are frequently obscured by numerous vaginal fluid components causing the clinician to miss the clue cells completely or to quantify them at levels below their actual level. This can result in a false negative test result. Therefore, it would be highly desirable to have a distinct anaslyte that is accurately and conveniently monitored and whose presence is correlated with clue cells.